ABSTRACT
The experimental data analysis demonstrates that it is possible to reconstruct hidden variables in a dynamical system directly from the observables by simultaneously reconstructing the equations of motion for the system. The trajectory method was developed for situations where all the variables required for an embedding of the dynamics can be measured. Unfortunately, this is almost never the case in real experiments. The experimentor frequently cannot measure one or more of the variables, i.e., some variables are hidden. Within the framework of the trajectory method, hidden variables and any preknowledge of the dynamics may be easily incorporated. The greatest advantage of the competing technique is that the resulting models can be interpreted physically so as to understand, as well as forecast, the dynamics. If the goal of modeling is forecasting alone, any approach is acceptable given that the predictions are accurate.
